基于MLR和LSTM神经网络的短期负荷预测方法

Short-term load prediction method based on MLR and LSTM neural networks

由于短期负荷非静止和强随机特征,难以准确预测负荷行为。为此,提出了改进的短期负荷预测方法。应用集合经验模态分解算法,依据频率从低到高将负荷分组;通过MLR预测平滑、周期的低频部分,保持高效的计算能力,而对具有强随机性的高频部分,则通过LSTM进行预测,即采用结合MLR和LSTM这两种方法获得实际预测负荷。最后,通过实验计算来自中国西部的测试数据,验证该方法的有效性。

Due to the non-stationary and strong stochastic characteristics of short-term load,it is difficult to accurately predict the load behavior.Therefore,an improved short-term load forecasting method is proposed.The load is grouped according to the frequency from low to high by ensemble empirical mode decomposition algorithm;MLR is used to predict the smooth and periodic low-frequency part to maintain efficient computing power,while LSTM is used to predict the high-frequency part with strong randomness,that is,MLR and LSTM are combined to obtain the actual predicted load;The effectiveness of the method is verified by the experimental calculation of the test data from Western China.